use of org.apache.flink.api.common.functions.FlatJoinFunction in project flink by apache.
the class InnerJoinOperatorBaseTest method testJoinPlain.
@Test
public void testJoinPlain() {
final FlatJoinFunction<String, String, Integer> joiner = new FlatJoinFunction<String, String, Integer>() {
@Override
public void join(String first, String second, Collector<Integer> out) throws Exception {
out.collect(first.length());
out.collect(second.length());
}
};
@SuppressWarnings({ "rawtypes", "unchecked" }) InnerJoinOperatorBase<String, String, Integer, FlatJoinFunction<String, String, Integer>> base = new InnerJoinOperatorBase(joiner, new BinaryOperatorInformation(BasicTypeInfo.STRING_TYPE_INFO, BasicTypeInfo.STRING_TYPE_INFO, BasicTypeInfo.INT_TYPE_INFO), new int[0], new int[0], "TestJoiner");
List<String> inputData1 = new ArrayList<String>(Arrays.asList("foo", "bar", "foobar"));
List<String> inputData2 = new ArrayList<String>(Arrays.asList("foobar", "foo"));
List<Integer> expected = new ArrayList<Integer>(Arrays.asList(3, 3, 6, 6));
try {
ExecutionConfig executionConfig = new ExecutionConfig();
executionConfig.disableObjectReuse();
List<Integer> resultSafe = base.executeOnCollections(inputData1, inputData2, null, executionConfig);
executionConfig.enableObjectReuse();
List<Integer> resultRegular = base.executeOnCollections(inputData1, inputData2, null, executionConfig);
assertEquals(expected, resultSafe);
assertEquals(expected, resultRegular);
} catch (Exception e) {
e.printStackTrace();
fail(e.getMessage());
}
}
use of org.apache.flink.api.common.functions.FlatJoinFunction in project flink by apache.
the class InnerJoinOperatorBaseTest method testTupleBaseJoiner.
@Test
public void testTupleBaseJoiner() {
final FlatJoinFunction<Tuple3<String, Double, Integer>, Tuple2<Integer, String>, Tuple2<Double, String>> joiner = new FlatJoinFunction<Tuple3<String, Double, Integer>, Tuple2<Integer, String>, Tuple2<Double, String>>() {
@Override
public void join(Tuple3<String, Double, Integer> first, Tuple2<Integer, String> second, Collector<Tuple2<Double, String>> out) {
assertEquals(first.f0, second.f1);
assertEquals(first.f2, second.f0);
out.collect(new Tuple2<>(first.f1, second.f0.toString()));
}
};
final TupleTypeInfo<Tuple3<String, Double, Integer>> leftTypeInfo = TupleTypeInfo.getBasicTupleTypeInfo(String.class, Double.class, Integer.class);
final TupleTypeInfo<Tuple2<Integer, String>> rightTypeInfo = TupleTypeInfo.getBasicTupleTypeInfo(Integer.class, String.class);
final TupleTypeInfo<Tuple2<Double, String>> outTypeInfo = TupleTypeInfo.getBasicTupleTypeInfo(Double.class, String.class);
final int[] leftKeys = new int[] { 0, 2 };
final int[] rightKeys = new int[] { 1, 0 };
final String taskName = "Collection based tuple joiner";
final BinaryOperatorInformation<Tuple3<String, Double, Integer>, Tuple2<Integer, String>, Tuple2<Double, String>> binaryOpInfo = new BinaryOperatorInformation<Tuple3<String, Double, Integer>, Tuple2<Integer, String>, Tuple2<Double, String>>(leftTypeInfo, rightTypeInfo, outTypeInfo);
final InnerJoinOperatorBase<Tuple3<String, Double, Integer>, Tuple2<Integer, String>, Tuple2<Double, String>, FlatJoinFunction<Tuple3<String, Double, Integer>, Tuple2<Integer, String>, Tuple2<Double, String>>> base = new InnerJoinOperatorBase<Tuple3<String, Double, Integer>, Tuple2<Integer, String>, Tuple2<Double, String>, FlatJoinFunction<Tuple3<String, Double, Integer>, Tuple2<Integer, String>, Tuple2<Double, String>>>(joiner, binaryOpInfo, leftKeys, rightKeys, taskName);
final List<Tuple3<String, Double, Integer>> inputData1 = new ArrayList<Tuple3<String, Double, Integer>>(Arrays.asList(new Tuple3<>("foo", 42.0, 1), new Tuple3<>("bar", 1.0, 2), new Tuple3<>("bar", 2.0, 3), new Tuple3<>("foobar", 3.0, 4), new Tuple3<>("bar", 3.0, 3)));
final List<Tuple2<Integer, String>> inputData2 = new ArrayList<Tuple2<Integer, String>>(Arrays.asList(new Tuple2<>(3, "bar"), new Tuple2<>(4, "foobar"), new Tuple2<>(2, "foo")));
final Set<Tuple2<Double, String>> expected = new HashSet<Tuple2<Double, String>>(Arrays.asList(new Tuple2<>(2.0, "3"), new Tuple2<>(3.0, "3"), new Tuple2<>(3.0, "4")));
try {
final TaskInfo taskInfo = new TaskInfo("op", 1, 0, 1, 0);
ExecutionConfig executionConfig = new ExecutionConfig();
executionConfig.disableObjectReuse();
List<Tuple2<Double, String>> resultSafe = base.executeOnCollections(inputData1, inputData2, new RuntimeUDFContext(taskInfo, null, executionConfig, new HashMap<String, Future<Path>>(), new HashMap<String, Accumulator<?, ?>>(), new UnregisteredMetricsGroup()), executionConfig);
executionConfig.enableObjectReuse();
List<Tuple2<Double, String>> resultRegular = base.executeOnCollections(inputData1, inputData2, new RuntimeUDFContext(taskInfo, null, executionConfig, new HashMap<String, Future<Path>>(), new HashMap<String, Accumulator<?, ?>>(), new UnregisteredMetricsGroup()), executionConfig);
assertEquals(expected, new HashSet<>(resultSafe));
assertEquals(expected, new HashSet<>(resultRegular));
} catch (Exception e) {
e.printStackTrace();
fail(e.getMessage());
}
}
use of org.apache.flink.api.common.functions.FlatJoinFunction in project flink by apache.
the class ReusingHashJoinIteratorITCase method testBuildSecond.
@Test
public void testBuildSecond() {
try {
TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, 500, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, 500, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
// collect expected data
final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = joinTuples(collectTupleData(input1), collectTupleData(input2));
final FlatJoinFunction matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();
// reset the generators
generator1.reset();
generator2.reset();
input1.reset();
input2.reset();
// compare with iterator values
ReusingBuildSecondHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator = new ReusingBuildSecondHashJoinIterator<>(input1, input2, this.recordSerializer, this.record1Comparator, this.recordSerializer, this.record2Comparator, this.recordPairComparator, this.memoryManager, ioManager, this.parentTask, 1.0, false, false, true);
iterator.open();
while (iterator.callWithNextKey(matcher, collector)) ;
iterator.close();
// assert that each expected match was seen
for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
if (!entry.getValue().isEmpty()) {
Assert.fail("Collection for key " + entry.getKey() + " is not empty");
}
}
} catch (Exception e) {
e.printStackTrace();
Assert.fail("An exception occurred during the test: " + e.getMessage());
}
}
use of org.apache.flink.api.common.functions.FlatJoinFunction in project flink by apache.
the class ReusingHashJoinIteratorITCase method testBuildSecondAndFullOuterJoin.
@Test
public void testBuildSecondAndFullOuterJoin() {
try {
TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, 1000, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, 500, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
// collect expected data
final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = fullOuterJoinTuples(collectTupleData(input1), collectTupleData(input2));
final FlatJoinFunction matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();
// reset the generators
generator1.reset();
generator2.reset();
input1.reset();
input2.reset();
// compare with iterator values
ReusingBuildSecondHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator = new ReusingBuildSecondHashJoinIterator<>(input1, input2, this.recordSerializer, this.record1Comparator, this.recordSerializer, this.record2Comparator, this.recordPairComparator, this.memoryManager, ioManager, this.parentTask, 1.0, true, true, false);
iterator.open();
while (iterator.callWithNextKey(matcher, collector)) ;
iterator.close();
// assert that each expected match was seen
for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
if (!entry.getValue().isEmpty()) {
Assert.fail("Collection for key " + entry.getKey() + " is not empty");
}
}
} catch (Exception e) {
e.printStackTrace();
Assert.fail("An exception occurred during the test: " + e.getMessage());
}
}
use of org.apache.flink.api.common.functions.FlatJoinFunction in project flink by apache.
the class ReusingHashJoinIteratorITCase method testBuildFirstAndBuildSideOuterJoin.
@Test
public void testBuildFirstAndBuildSideOuterJoin() {
try {
TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, 500, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, 1000, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
// collect expected data
final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = leftOuterJoinTuples(collectTupleData(input1), collectTupleData(input2));
final FlatJoinFunction matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();
// reset the generators
generator1.reset();
generator2.reset();
input1.reset();
input2.reset();
// compare with iterator values
ReusingBuildFirstHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator = new ReusingBuildFirstHashJoinIterator<>(input1, input2, this.recordSerializer, this.record1Comparator, this.recordSerializer, this.record2Comparator, this.recordPairComparator, this.memoryManager, ioManager, this.parentTask, 1.0, false, true, false);
iterator.open();
while (iterator.callWithNextKey(matcher, collector)) ;
iterator.close();
// assert that each expected match was seen
for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
if (!entry.getValue().isEmpty()) {
Assert.fail("Collection for key " + entry.getKey() + " is not empty");
}
}
} catch (Exception e) {
e.printStackTrace();
Assert.fail("An exception occurred during the test: " + e.getMessage());
}
}
Aggregations